The foundation of PowerFlex is its software-first design. At the base, it is a software-defined block-storage service enabling a scale-out storage infrastructure using x86 nodes and TCP/IP networking. PowerFlex software runs on and integrates with a large number of hypervisors and operating systems.

PowerFlex provides flexible consumption options on the hardware infrastructure side. Nodes can be configured with any number of component options to meet varying customer needs.

PowerFlex rack is a fully engineered rack-scale system with integrated networking and intelligent cabinet design. The software and hardware components of a PowerFlex rack are managed through PowerFlex Manager, a comprehensive management stack for all layers of the deployment. PowerFlex rack is designed to simplify deployments and accelerate time-to-value.

PowerFlex appliance is flexible node-oriented option and provides a modest starting point with a massive scale potential. It allows you to utilize a broader set of networking options, either fully or partially managed. Like the rack, PowerFlex appliance supports the full set of PowerFlex functionality, and it is also lifecycle managed with PowerFlex Manager.

PowerFlex custom nodes are the name for the R650 and R750-based nodes that are not managed by PowerFlex Manager. (Note: the older R640, R740, R840-based VxFlex Ready Nodes were not rebranded to PowerFlex.) PowerFlex custom nodes provide an extremely flexible foundation with fully tested and optimized servers. Management of custom nodes is done through the individual management tools of each system component.

Main use cases for the Flex family:

  • Two-Layer Architecture and storage-only options: Customers can redesign their storage environment using PowerFlex storage software to replicate a traditional two-layer SAN architecture model. A PowerFlex two-layer deployment provides efficient parallelism and no single points of failure. Storage and compute nodes remain separate operationally, giving teams the flexibility to manage each infrastructure independently.
  • Multi-hypervisor and Bare Metal Support: The PowerFlex family offers VMware® vSphere integration alongside the ability to support other hypervisors (KVM, Hyper-V) and even bare-metal configurations. This unique ability to work with hypervisors or standard operating systems provides workload flexibility and gives groups within the organization the ability to change requirements as needed if new projects and workloads arise without lock-in. Storage pools are not limited by ESXi cluster sizing and may span VMware clusters.
  • High Performance for Applications and Databases: Consolidate multiple workloads that demand high performance. PowerFlex delivers consistent sub 1 MS response times and removes I/O bottlenecks. IOPS scale linearly while latency remains at 1 MS as nodes are added. Scale seamlessly and add capacity on the fly. Enterprise-grade data protection, support for multi-tenant platform architectures, and add-on features. Massively parallel drive and node rebuilds provide quick remediation if hardware fails. Resiliency scales linearly with cluster expansion.

Features that are unique to the PowerFlex family:

Flexibility – The PowerFlex family emphasizes the word “flex” indicating it accommodates a wide range of flexible options. The PowerFlex family offers flexibility in multiple ways:

  • There’s the architecture which supports multi-OS, multi-hypervisor and bare metal options in the same cluster (Linux or Windows; ESXi, RHV/KVM, or Hyper-V)
  • Customers have flexibility in deployment options and can choose 2-layer or HCI, compute or storage only, or mix and match
    • Notably, deploying as 2-layer does not incur a performance penalty, as with other software-based systems. PowerFlex achieves performance without requiring data locality
    • PowerFlex uses a unique protocol to move storage commands over the TCP/IP layer. It’s efficient, lightweight, and massively parallel
  • Support for multiple types of Flash, including NVMe and future storage class memory
  • And there is workload flexibility with the ability to run traditional workloads as well as new scale-out workloads

Linear Scalability and Elasticity – A common misconception is that the PowerFlex family is for large, enterprise accounts only. This is not true. The PowerFlex family enables customers of all sizes to support their (growing and changing) environment with its scalability and elasticity:

  • Start small and grow incrementally with no bottlenecks or resiliency tradeoffs
  • Scale compute and storage independently (or together) for minimum TCO
  • Deliver consistent performance and latency with linear scalability
  • Easily adjust/grow the system at the application consumption level without disruption to running apps.

Predictable High Performance and Resiliency – This builds on the scale and elasticity. The PowerFlex family provides unmatched performance and resiliency in the following ways:

  • Multiple protection domains in a single cluster
  • Designed for extensive fault tolerance and high availability (no single point of failure)
  • Reliable, repeatable, fast rebuilds deliver 6×9’s Tier 1 Resiliency
  • Predictability even in mixed workloads with high variability
    • Performance grows linearly as you scale, and so does resiliency, with rebuild speed and efficiency also increasing in near-linear fashion
    • Example: Failed 750GB SSD in a storage pool with 14 contributing nodes, required rebuilding 344GB of data. Rebuild time = 75 seconds
    • Example: Failed node in SSD pool required rebuild of 2.1TB data; 13 nodes contributing to rebuild. Time to complete rebuild = 261 seconds (4m 21s)

Full Stack Architecture – another unique benefit of the Flex family is its M&O and single source of support:

  • Ability to take M&O to the network level
  • Full system optimization: optimize reallocated storage, low system overhead with extremely small footprint
  • Single source of support for both hardware and software
    • *Proof points and claims are compared to a manual approach without PowerFlex Manager

PowerFlex is designed to operate as a mesh. Although storage gets logically subdivided, in every case, whether converged or 2-layer we’re going to see a data client communicating with multiple data servers (SDS), even if it only has a single volume provisioned from the PowerFlex storage pool.

The example here is a 2-layer topology.( The connecting lines for an HCI mesh would be much more complex to visualize. )

The approach is to maximize throughput and minimize round-trip times by distributing the load across all of the storage devices present. This wouldn’t work unless PowerFlex also worked to distribute the data evenly across all of the storage devices present, and the next section will deal with exactly how we accomplish that with data layout and maintain that balance as the storage system takes on more data.

A big release which unifies PowerFlex management and orchestration across the entire offering including SW customers, and adds new client connectivity and storage protocols.

  • PowerFlex 4.0
    • NVMe over Fabric (TCP) – Add NVMe/TCP as a frontend protocol in addition to current SDC (agent-based) access.
      • No agent needed for connectivity
    • SDNAS (File) – Software-Defined File Services  (NFS/SMB) as supplement to the Block Storage services in PowerFlex
      • Performance and Scale: Transactional databases and traditional NAS workloads
      • High Availability: 30 seconds target
      • Fault Tolerance & Resiliency: LACP Bonds
      • File Maintenance mode: New maintenance mode for serviceability operation
      • Management and Monitoring: UI and REST API, CloudIQ Integration, SNMP v2/v3 support
    • PowerFlex Manager
    • Unified Manager – PowerFlex will provide a unified interface and full LCM services for improved user experience, manageability, scalability, resiliency. Enhancements will include Role & Scope based Access Controls, Enhanced events, alerts and historical metrics, enable a new common PowerAPIs to improve ecosystem and customer integration.
    • Unified catalogs – Provide a common compliance validation and upgrade management across PowerFlex appliance and rack. From this release on, appliance will share a common RCM base with rack. appliance installs can get compliance results from an RCM catalog, the upgrade matrix will show appliance IC to RCM as a valid upgrade (appliance IC will transition to RCM).
    • Remove vSAN dependency – Remove the dependency on vSAN as the shared storage infrastructure for the PowerFlex management cluster
    • Modernized serviceability with ESE (Dreamcatcher aka Embedded SupportAssist Enabler v5). SCG (SAE 5.0) – Originally planned for Hyacinth but might be delivered in Jade, PowerFlex family should use SAE 5 aka SCG as the replacement for the SRS gateway. This is to keep the customer and our system running on the most current and secure connectivity tool Dell uses. This should include both greenfield and brownfield deployments
  • PowerFlex Platform
    • Security/Fed– VPAT, FIPS HW Level 2 cert, STIG
    • Cisco 96 Port Switch 93360YC-FX2
  • Ecosystem
    • CSM Container Storage Modules (Karavi) Tech preview

Replication – Exposes asynchronous replication capability to be used with K8s

PowerFlex File – Design Principles

  • NAS nodes
    • Dedicated nodes to host the NAS containers
    • PowerFlex Appliance and Rack to use PowerEdge R650-based NAS nodes
  • NAS cluster
    • There will be one NAS cluster per PowerFlex system (MDM-cluster)
    • An NAS cluster can have 2-16 physical NAS nodes
  • Storage Pools
    • Every Protection Domain will have a Storage Pool dedicated for NAS
    • SP to house NAS configuration volumes for “NAS Servers”
      • SP can also be used for user filesystems
      • NAS Server’
    • A logical construct that provides data segregation
      • A “NAS Server” hosts multiple File Systems
      • NAS Server configuration volumes will be thin provisioned
  • File Systems
    • Bound to it’s parent “NAS Server”
    • Served by a single, dedicated PowerFlex (block) volume and is transparent to the user
    • User file systems benefit from the full performance and data resiliency of PowerFlex block volumes
  • Supported protocols
    • NFS and SMB support (NFSv3 and v4, SMBv2 and v3)
    • FTP and SFTP
  • File System Operations
    • User quotas and Tree Quotas
    • Extend/shrink file system (space reclaim)
    • File system read/write snapshots
  • Workloads
    • Transactional databases and Traditional NAS workloads
  • Data Reduction

Inline Compression

  • Management and Monitoring
    • UI and REST API
    • CloudIQ Integration
    • SNMP v2 and v3 support
  • Data Protection
    • 3-way NDMP support for backup
  • Security
    • CAVA (Common Antivirus Agent for SMB clients)
    • D@RE with CloudLink
  • Serviceability
    • SRS/ESE (Call Home)
    • Alerts
    • Data collection aka “native audit log”

PowerFlex 4.0 NVMe/TCP

  • NVMe was designed for direct attached PCIe SSDs
    • Expanded to NVMe over fabrics (NVMe-oF), using simple and efficient TCP/IP fabric
    • TCP transport provides a standard and interoperable solution
    • Allows the use of existing network and hardware
    • Hosts can use standard NVMe/TCP OS drivers already available in mainstream Linux distributions and in ESXi
    • NVMe/TCP is an emerging technology
    • Ecosystem is still evolving and catching up with Fiber Channel

PowerFlex – Transition to NVMe/TCP Host Connectivity (Front-End)


NVMe/TCP Deployment

  • Deployed using PowerFlex Manager (storage nodes)
  • 2-layer deployment (HCI support to follow)
  • Supports mainstream Linux and ESXi
    • ESXi 7.0U3, RHEL 8.6 and later, SLES 15SP3, Ubuntu 20.04
  • SDC-free storage connectivity
    • Doesn’t require proprietary kernel drivers
  • Co-exists with PowerFlex SDC
    • SDC and NVMe/TCP can share the same system (can’t share same volumes)

SDC connectivity still outperforms NVMe/TCP

PowerFlex IO Path

PowerFlex 4.0 – Unified Management


  • Installation Manager and Gateway (IM-GW)
    • PowerFlex core SW deployment and upgrade
    • REST API for PowerFlex core SW
    • Self RBAC
    • Alerts for PowerFlex storage
    • SNMP, SRS, CloudIQ
  • Presentation Server
    • WebUI for PowerFlex core SW
    • Self RBAC
    • Alerts for PowerFlex core
  • PowerFlex Manager (PFxM)
    • Rack and appliance Deployment and LCM incl. nodes and supported switches
    • REST API for PowerFlex Manager
    • Self RBAC
    • HW/network alerts (uses IM-GW polling PowerFlex core SW)
    • SNMP, SRS, CloudIQ


  • PowerFlex manager (UX, API, CLI)
    • Unified WebUI
      • REST API for rack, appliance and Powerflex core SW
      • Block and File services
      • SDC and NVMe/TCP hosts
    • Single REST endpoint for all APIs
      • Legacy and modern, ISG-standard PowerAPI
  • Unified services:
    • Deployment and LCM
    • Unified RBAC and user management
      • No separate users or LDAP configuration
      • Single set of roles
    • Security
    • Events, Alerts and reporting
    • Logging and metrics collection
    • SNMP, SRS, SMTP, CloudIQ
  • Resilient management stack (HA)
  • Modular and extendible

Below, you can see a screenshot, showing the new, 4.0 dashboard

So, Why PowerFlex?

1 | Unbounded software-defined infrastructure for unconstrained consolidation

PowerFlex is a powerful platform for broad consolidation initiatives. PowerFlex unifies the delivery of block storage, file storage and compute resources in an engineered integrated system, while supporting a wide range of hypervisors and operating systems on a common platform. PowerFlex also utilizes standards based NVMe/TCP connectivity. These capabilities enable organizations to optimally consolidate a diverse set of workloads with varying requirements and operating environments.

2 | Adaptable architecture boosts IT responsiveness

PowerFlex scales compute and storage resources flexibly and non-disruptively to address specific bottlenecks. Rapidly compose and recompose resource pools to address changing business needs. Break through traditional limitations, scale your infrastructure, and grow and shrink resource pools as needs dictate, without complex planning.

3 | Smart optimization ensures extreme performance and scalability

PowerFlex software driven optimization consolidates resources across nodes while optimizing data path and placement, resulting in unrestricted IO and throughput performance with sub-millisecond latency. The architecture can linearly scale to 1000s of nodes, ensuring optimal outcomes whether you are modernizing a high-performance database or consolidating large application landscapes.

4 | Self-healing capabilities deliver uninterrupted services

PowerFlex systems are designed to sustain multiple subsequent failures without service outages or performance degradation. PowerFlex continuously monitors resources and rapidly rebuilds datasets residing on failed drives or nodes, ensuring uninterrupted service availability while delivering compliance with performance targets.

5 | Intelligent insights and unified management simplify operations at scale

PowerFlex Manager, a unified management toolset for PowerFlex systems, is designed to simplify IT Operations and lifecycle management tasks by providing extensive automation that reduces administrative time by up to 95% and the number of steps involved in operations by over 77%. Integration with CloudIQ intelligent insights further simplifies monitoring of distributed multi-location PowerFlex deployments by providing a seamless cloud-based AIOps mechanism.

6 | Broad hyperscaler support maximizes multi-cloud agility and flexibility

PowerFlex offers choice and freedom when it comes to multi-cloud deployments. With support for all major hyperscaler platforms, including AWS EKS Anywhere, Google Cloud Anthos, and Azure Arc-enabled Kubernetes, you can simplify your multi-cloud operation across multiple hyperscaler cloud platforms with a unified on-premises PowerFlex infrastructure.

7 | Comprehensive tools enhance DevOps productivity and IT agility

In addition to hyperscaler support, PowerFlex supports a wide range of on-premises container orchestration platforms on bare metal or hypervisors including Red Hat OpenShift, SUSE Rancher, and VMware Tanzu. Through Dell Container Storage Modules (CSM) and CSI drivers, PowerFlex simplifies storage operations for Kubernetes. Additionally, PowerFlex offers a rich and out-of-the-box toolset that includes PowerFlex REST APIs, PowerFlex Ansible modules, and Dell AppSync to automate DevOps, infrastructure management, and DBA workflows.

8 | Rich validated ecosystem optimizes workload execution

PowerFlex is optimized and validated for a broad set of enterprise workloads to deliver exceptional performance outcomes. PowerFlex provides industry leading performance and predictability for IO-intensive transactional or throughput-intensive analytics workloads while minimizing hardware footprint.

9 | Enterprise data services simplify protection and compliance

PowerFlex provides enterprise data services that include compression, encryption, snapshots and replication to help meet specific workload and corporate requirements. The PowerFlex solution with Dell PowerProtect Cyber Recovery further enhances the resiliency and protection of these mission-critical IT assets.

10 | Dell Technologies – Your trusted partner for long-term success

PowerFlex is engineered, deployed, and supported as one system by Dell Technologies, minimizing risks in implementations and delivering predictable outcomes. PowerFlex customers benefit from Dell offerings in the following areas:

-Flexible on-demand OPEX consumption with APEX Flex on Demand and APEX Datacenter Utility

-Accelerated enterprise technology adoption with Dell ProDeploy Enterprise Suite

-Comprehensive support for complex environments with Dell ProSupport Enterprise Suite and Optional Support Services for Enterprise

-Maximize investments, safeguard the future and deliver guaranteed outcomes with Dell’s Future-Proof Program

Leave a Reply